A beverage manufacturer partnering with a retailer wanted to reduce the in-store out-of-stock rate for the manufacturer's products.
The daily historical series of product sales from dozens of stores was processed and characterization attributes were added to the SKUs.
BOTs were added to the Forecast models that generate the best order based on stock level, order flow and service level.
The global prediction error (MAPE) obtained was 10%. The models, together with the ordering BOTs, promoted a 50% reduction in outage events, without increasing inventory levels and with a consequent increase in revenue.